Intel Gaudi Software Stack Verification

This section provides the required instructions to verify the Intel® Gaudi® software already installed on the system. Start by determining what is already installed on the platform, then decide the next steps for installation.

Note

Refer to the Support Matrix for the list of supported and validated FW and SW components. The versions listed in the Support Matrix are the validated configurations from Intel Gaudi.

Configurations not tested by Intel Gaudi may not perform or function properly.

Check Existing Software Installation

Follow the below steps to check the existing SW and FW components on the platform.

1. To verify the current Intel Gaudi SW version, run the hl-smi tool. Use the HL-SMI Version at the top of the output. For example, if the installed version is 1.16.1, the output should be as follows:

      HL-SMI Version:       hl-1.16.1-XXXXXXX
      Driver Version:       1.16.1-XXXXXX
   

2. To verify the system SW, run the following command:

   apt list --installed | grep habana
   

   If the installation was successful, you should see the habanalabs-dkms driver and the following SW components at the root system:

    habanalabs-container-runtime/focal,now 1.16.1-7 amd64 [installed]
    habanalabs-dkms/focal,focal,now 1.16.1-7 all [installed]
    habanalabs-firmware/focal,now 1.16.1-7 amd64 [installed]
    habanalabs-firmware-odm/focal,now 1.16.1-7 amd64 [installed]
    habanalabs-firmware-tools/focal,now 1.16.1-7 amd64 [installed]
    habanalabs-graph/focal,now 1.16.1-7 amd64 [installed]
    habanalabs-rdma-core/now 1.16.1-7 all [installed]+
    habanalabs-qual/focal,now 1.16.1-7 amd64 [installed]
    habanalabs-thunk/focal,focal,now 1.16.1-7 all [installed]
   

   Note

   • Only the habanalabs-firmware-tools, habanalabs-graph, habanalabs-rdma-core, and habanalabs-thunk packages are visible in the Intel Gaudi Docker image.

   • The habanalabs-rdma-core package is installed on Gaudi 2 only.

3. To verify the PyTorch framework used on the system, run the following command:

        pip list | grep habana
   
        habana-media-loader     1.16.1-7
        habana-pyhlml           1.16.1-7
        habana-torch-dataloader 1.16.1-7
        habana-torch-plugin     1.16.1-7
        lightning-habana        1.4.0
   

Next Steps

After verifying the installed versions, you can do the following:

• Install the Intel Gaudi SW stack and PyTorch framework on a bare metal platform using the habanalabs-installer.sh script as detailed in the sw_stack_installation_bare and native_frameworks_installation_bare sections.

• Or, install using a container. Make sure the Intel Gaudi SW stack driver is installed and then follow the instructions to prebuilt_containers_bare. Using Intel Gaudi Docker images is recommended.